K A R A B E C K E R , R E E D C O L L E G E S A M E E R U D D O W L A K H A N , R E E D C O L L E G E

L A L Z I M M A N , U C S A N T A B A R B A R A

G U E S T L E C T U R E B Y S A M E E R U D D O W L A K H A N F O R C H R I S T I N A M . E S P O S I T O

1 0 F E B R U A R Y 2 0 1 6

Creaky Voice in a diverse gender sample: Challenging ideologies about sex, gender

and creak in American English

Background

•  Conflicting perspectives on the link between gender and creaky voice, which has been associated with: •  Men in the UK (Esling 1978; Henton & Bladon 1988; Stuart-Smith 1999)

•  Young, white women in the US (Yuasa 2010; Podesva 2013)

•  Chicano/a gangsters (Mendoza-Denton 2007, 2011)

•  Men who are (perceived as) gay/queer (Podesva 2007; Zimman 2013)

•  Gendered meaning of creak is still uncertain

Background

•  Previous studies all operate within the gender binary •  Limitations of this view: •  Restricts our theorization of gendered social meaning to two

options: maleness/masculinity and femaleness/femininity •  Encourages simplification of gendered meaning based on

broad correlations •  Does not reflect the full diversity of the population. What

about trans speakers? •  Hard to tease apart socialization, identity, physiology

Our goals

•  To include speakers with a wider range of identifications related to sex and gender

•  To take a more nuanced view of gender identity and different aspects of sex

•  To pay special attention to how creak relates to queer and trans identities •  Zimman (2012, 2013) suggests that trans men may be

especially creaky

Who uses creak? Can a more diverse sample wrt sex and gender help us get a handle on creak’s social meanings?

Coding for sex and gender

•  Factors to consider •  Identity: current self-reported gender •  Socialization: sex assignment at birth •  Laryngeal physiology: exposure to testosterone

•  Operationalized as 2 factors, each with 3 levels •  Gender identity: female, male, non-binary •  Sex assigned at birth and current hormonal status:

•  Male-assigned (AMAB) •  Female-assigned, not on testosterone (AFAB) •  Female-assigned, on testosterone (AFAB+T)

Identity categories

Female Male Non-binary

AMAB Trans women Cis men AMAB non-binary

AFAB Cis women Trans men, not on testosterone

AFAB non-binary, not on testosterone

AFAB+T N/A Trans men, on testosterone

AFAB non-binary, on testosterone

Methods: Data collection

•  Native speakers of American English, ages 18-35: •  Recorded in 2013, at Reed College Lab of Linguistics (LoL) in

Portland, Oregon •  80% self-reported as white or Caucasian

•  2 speech styles: •  Casual interview discussing speaker’s hometown •  Scripted reading of the Rainbow Passage

Methods: Speakers

•  Speakers self-reported their gender identity, assigned sex, and hormonal status in a post-recording questionnaire and/or online survey

Female Male Non-binary Total

AMAB 8 6 5 19 AFAB 6 2 10 18 AFAB+T N/A 7 7 14 Total 14 15 22 51

Methods: Subsample

We present results from a subsample of 27 speakers:

Female Male Non-binary Total

AMAB 3 6 3 12 AFAB 6 2 1 9 AFAB+T N/A 4 2 6 Total 9 12 6 27

Methods: Transcription

•  Segmental: Phonemic transcriptions and segmentations automatically generated by FAVE (Rosenfelder et al. 2011)

•  Full recording of Rainbow Passage •  First 5 minutes of the casual interview recording •  Boundaries hand-corrected by RA in cases of FAVE error

•  Prosodic: Locations of right edges of intonation phrases (IPs), and IP boundary tone type, using ToBI labels (Veilleux et al. 2006, Beckman & Ayers Elam 1997) •  L-L%: low falling (e.g. declarative) •  L-H%: low rising (e.g. continuation rise) •  H-H%: high rising (e.g. yes/no question) •  H-L%: high plateau (e.g. trailing off) •  !H-L%: downstepped plateau (e.g. calling contour)

Methods: Auditory coding

•  Each vowel was coded by a primary coder for: •  Voice quality: modal, creaky, other (e.g. breathy), or unsure • 

• 

•  A secondary coder coded for voice quality •  Mean inter-coder reliability across recordings: 84.9%

•  Discrepancies between coders resolved by authors •  In cases where no 2 coders agreed on a given vowel’s voice

quality code, that vowel was discarded

Methods: Auditory coding

•  All coding was done auditorily, as in previous work on creak and gender (Podesva 2013; Zimman 2013)

•  Choice also based on results of related work (Khan et al. 2015) •  16 trained linguistics students rated relative creakiness of phrase-final

word ‘bows’ extracted from the Rainbow Passage •  Crosslinguistic acoustic cues for creak (H1-H2, H1-A1, H1-A3) were

found not to be correlated with our students’ creakiness ratings •  Creaky voice in American English, at least, seems to be cued by a more

complex array of acoustic features

The data: overall patterns

� Wide range across speakers in use of creak

0

10

20

30

40

50

60

70

C37

C21

C62

C57

C32

C15

C60

C63

Z16 Z5

C2

4 C5

8 Z3

6 C2

7 C3

8 C6

C36

Z28

C43

C39

C67

C25 Z3

C4

4 C3

3 C5

5 C2

8

Casual Scripted

The data: overall patterns

Female Male Non-Binary total

AMAB 26% 26% 27% 26%

AFAB 38% 35% 22% 37%

AFAB+T n/a 23% 23% 23%

total 35% 26% 29%

Modeling

�  We ran a mixed-model logistic regression on individual vowels ¡  Response: Voice quality (creak is the application value) ¡  Predictors:

Internal factors •  Word (random) •  Vowel quality •  Stress (primary, secondary, none) •  Pitch accented (yes, no) •  Position in IP (final, non-final) •  IP bndry tone (LL, LH, HH, HL) •  IP-initial vowel (yes, no) •  Style (casual, reading)

Social factors •  Speaker (random) •  Year of birth (continuous) •  Gender id (F, M, non-binary) •  Sex (AFAB, AFAB+T, AMAB) •  Sexual orientation (asexual, bi,

gay, queer, straight)

Results: Internal factors

Factor p value Level n Proportion of vowels that are creaky

Factor weight

IP boundary tone p < .001 …

Position in IP p < .001 …

IP-initial vowel p < .001 …

Vowel quality p < .001 …

Stress p = .00161 …

Style p < .001 Casual 0.248 13788 .31 .562

Reading -0.248 7054 .25 .438

� All internal factors selected were in line with previous literature

Results: Social factors

Factor p value Level n Proportion of vowels that are creaky

Factor weight

Sex p = .0419 AFAB 0.492 7126 .37 .621

AMAB -0.239 9230 .26 .441

AFAB + T -0.254 4486 .23 .437

� The social factors show a more complicated picture ¡  Gender identity, sexual orientation, and age are NOT

predictors of creak as a main effect ¡  Sex is the only social predictor of creak

Discussion

•  We return to our original goals, specifically in how we tease apart various aspects of sex and gender. •  Identity: self-reported gender identity •  Socialization: sex assignment at birth •  Laryngeal physiology: exposure to testosterone

•  We’ve determined that gender identity does not predict use of creak

Discussion

•  So should we just replace “gender” with “sex assignment at birth”?

•  Not exactly. Sex assignment at birth alone does not predict creak either.

•  Speakers assigned female at birth fall into two categories with respect to use of creak: •  AFAB speakers (not on T) favor use of creak •  AFAB+T speakers disfavor use of creak, like AMAB speakers

Discussion

•  Ah! So then is it the physiological aspect of sex? •  Laryngeal physiology is the common trait here. •  AFAB speakers not on T favor creak •  AMAB and AFAB+T speakers disfavor creak

•  Helps us group cis men and trans men on T (less creak) apart from trans men not on T (more creak).

Discussion

•  One possible interpretation: •  Men are presumably motivated to sound more masculine •  Cis men and trans men on T achieve this through laryngeal

changes from testosterone exposure •  Trans men not on T may utilize creak as an alternative

•  However, this account cannot explain why cis women use creak at such a high rate! •  We do not assume that all people, regardless of gender or sex,

are motivated to sound more masculine •  Thus, this cannot be the end of the story

Conclusions

�  Our study complicates our view of gender and creak ¡  Wider range of sex/gender categories ¡  Explicitly tease apart separate influences of identity,

socialization, and physiology

�  Laryngeal physiology was the best predictor of creak use, not gender identity or assignment at birth

�  We’re currently working to further complicate this finding with additional data in our sample

�  We invite your input and suggestions

Acknowledgments

•  This project was funded by Reed College’s Stillman Drake Fund and Summer Scholarship Fund.

•  Many thanks to our speakers, our tireless RAs (pictured below), and to you at Macalester College!

References Beckman, Mary; Ayers Elam, Gayle. 1997. Guidelines for ToBI labeling, Version 3. Ohio State University ms. Henton, Caroline; Bladon, Anthony. 1988. Creak as a sociophonetic marker. In Hyman, Larry; Li, Charles N. (eds.) Language, Speech, and Mind. Longon: Routledge, pp 3-29. Khan, Sameer ud Dowla; Becker, Kara; Zimman, Lal. 2015. Acoustic correlates of creaky voice in English. Talk presented at the 170th Meeting of the Acoustical Society of America, Jacksonville. Mendoza-Denton, Norma. 2007. Creaky voice in gang girl narratives. Presented at the Annual Meetings of the American Anthropological Association, Washington, DC. Mendoza-Denton, Norma. 2011. The semiotic hitchhiker's guide to creaky voice: Circulation and gendered hardcore in a Chicana/o gang persona. Journal of Linguistic Anthropology 21 (2): 261-280. Podesva, Robert J. 2007. Phonation type as a stylistic variable: The use of falsetto in constructing a persona. Journal of Sociolinguistics 11: 478-504. Podesva, Robert J. 2013. Gender and the social meaning of non-modal phonation types. Proceedings of the 37th Annual Meeting of the Berkeley Linguistics Society 427-448. Rosenfelder, Ingrid; Fruehwald, Joe; Evanini, Keelan; Yuan, Jiahong. 2011. FAVE (Forced Alignment and Vowel Extraction) Program Suite. http://fave.ling.upenn.edu. Stuart-Smith, Jane. 1999. Voice quality in Glaswegian. Proceedings of the International Congress of Phonetic Sciences 14: 2553-2556. Yuasa, Ikuko Patricia. 2010. Creaky voice: A new feminine voice quality for young urban-oriented upwardly mobile American women? American Speech 85: 315-337. Veilleux, Nanette; Shattuck-Hufnagel, Stefanie; Brugos, Alejna. 2006. Transcribing Prosodic Structure of Spoken Utterances with ToBI. MIT OpenCourseWare. http://ocw.mit.edu Zimman, Lal. 2012. Voices in transition: Testosterone, transmasculinity, and the gendered voice among female-to-male transgender people. PhD dissertation, University of Colorado Boulder. Zimman, Lal. 2013. Hegemonic masculinity and the variability of gay-sounding speech: The perceived sexuality of transgender men. Journal of Language & Sexuality 2(1): 5-43.